feat(benchmarking): add mixed workload benchmark (TestMixedTraffic) #3186
Description
Summary
Add a benchmark test that runs multiple spamoor scenarios simultaneously against the same chain to simulate realistic L2 traffic patterns. Currently each benchmark test isolates a single workload type, which measures that workload ceiling but does not represent production traffic mixes.
Motivation
The DeFi simulation benchmark (TestDeFiSimulation) is bottlenecked by spamoor uniswap-swaps scenario - each swap requires a synchronous eth_call to GetAmountsIn/GetAmountsOut before sending, limiting injection to ~150-300 TPS regardless of config. A mixed workload avoids this single-scenario bottleneck while producing more realistic throughput numbers.
Proposed workload mix
| Spammers | Scenario | Gas/tx | Role |
|---|---|---|---|
| 2 | eoatx | ~21K | simple transfers, high TPS filler |
| 2 | erc20tx | ~65K | storage-heavy contract calls |
| 1 | uniswap-swaps | ~200K | DeFi call chains |
| 1 | gasburnertx | configurable | compute pressure |
This mix exercises simple transfers, storage reads/writes, deep call chains, and compute simultaneously - closer to what a production L2 mempool looks like.
Acceptance criteria
- New TestMixedTraffic in test/e2e/benchmark/
- Uses benchConfig env vars for system config (block time, gas limit, etc.)
- All spammer types run concurrently against the same chain
- Measurement window excludes warmup from all scenarios
- Results include per-scenario and aggregate metrics
Related
- Epic: [EPIC] Benchmarks #2288
- DeFi benchmark: cian/add-defi-simulation-benchmark
- Spamoor bottleneck analysis: see PR for TestDeFiSimulation